Multi-Armed Bandits with Fairness Constraints for Distributing Resources to Human Teammates
This addresses fairness in human-robot collaboration, specifically for resource distribution, but is incremental as it applies known bandit methods with fairness constraints.
The paper tackles the problem of a robot distributing resources to human teammates with unknown skill levels, introducing a multi-armed bandit algorithm with fairness constraints to ensure minimum selection rates. Results from a user study show that fairness significantly affects users' trust in the system.
How should a robot that collaborates with multiple people decide upon the distribution of resources (e.g. social attention, or parts needed for an assembly)? People are uniquely attuned to how resources are distributed. A decision to distribute more resources to one team member than another might be perceived as unfair with potentially detrimental effects for trust. We introduce a multi-armed bandit algorithm with fairness constraints, where a robot distributes resources to human teammates of different skill levels. In this problem, the robot does not know the skill level of each human teammate, but learns it by observing their performance over time. We define fairness as a constraint on the minimum rate that each human teammate is selected throughout the task. We provide theoretical guarantees on performance and perform a large-scale user study, where we adjust the level of fairness in our algorithm. Results show that fairness in resource distribution has a significant effect on users' trust in the system.